# hanruihua/NeuPAN

[TRO 2025] NeuPAN: Direct Point Robot Navigation with End-to-End Model-based Learning.

Repository: https://github.com/hanruihua/NeuPAN
Canonical: https://ross.abutalabs.com/products/neupan
Homepage: https://hanruihua.github.io/neupan_project/
Language: Python
License: GPL-3.0
License Family: copyleft
Topics: collision-avoidance, end-to-end, model-based-learning, robot-navigation, motion-planning, planning, planner, ir-sim, local-planner, mobile-robot-navigation, path-planning
Last push: 2026-06-21T14:03:12+00:00

## Health v2 (maintenance only)
Score: 69/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 88, release rhythm 45, longevity 67
- inputs: {"age_days": 939, "days_push": 73, "days_rel": 209, "gap_med": 157.0, "n_releases_24m": 3}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1090, forks 105 (observed 2026-08-28T04:03:32.794728+00:00)

## What it is
NeuPAN is a real-time, map-free robot motion planner that maps raw obstacle point data directly to control actions using an end-to-end MPC-based neural proximal alternating-minimization network. It is a Python library for point-robot navigation with collision avoidance, deployable on mobile robots and vehicles, with ROS/ROS2 wrappers available.

## Use cases
- navigate a mobile robot through cluttered unknown environments
- real-time collision avoidance from raw point cloud data
- local motion planning for nonholonomic robots
- deploy a model-based learning planner on a real robot
- plan paths for wheel-legged robots and autonomous vehicles
- integrate a planner into ROS or ROS2 navigation stacks

## When to choose
- you need real-time, map-free local planning with point-level collision avoidance
- you want an interpretable end-to-end planner combining learning and optimization
- you deploy on ground robots, wheel-legged robots, or autonomous vehicles
- you want a planner that integrates with ROS/ROS2

## When to avoid
- you need global path planning over large known maps
- you work outside Python or need non-GPL-licensed code
- you need planners for aerial or manipulator robots
- you require a fully mature production-grade navigation stack

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, simulation, robotics
- domain: robotics, autonomous-vehicles, machine-learning, simulation
- platform: python, cross-platform
- tags: motion-planning, collision-avoidance, mpc, model-based-learning, local-planner, mobile-robot, point-cloud, ros, end-to-end-learning, navigation, linux

## Member repositories
- hanruihua/NeuPAN (main) score 69

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:32.794728+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T06:49:13.246407+00:00, confidence not recorded.
  - readme: https://github.com/hanruihua/NeuPAN (fetched 2026-08-28T04:03:32.794728+00:00, sha e8fb6139c4c7)
  - homepage: https://hanruihua.github.io/neupan_project/ (fetched 2026-08-29T12:51:37.736010+00:00, sha b3370ceb5093)
- Data as of 2026-08-30T08:39:29.467469+00:00.